Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
MIME-Version: 1.0
References: <20220905045341.66191-1-jasowang@redhat.com> <20220905031405-mutt-send-email-mst@kernel.org>
 <CACGkMEtjQ0Jfok-gcRW+kuinsua2X0TscyTNfBJoXHny0Yob+g@mail.gmail.com>
 <056ba905a2579903a372258383afdf6579767ad0.camel@redhat.com>
 <CACGkMEuiDqqOEKUWRN9LvQKv8Jz4mi3aSZMwbhUsJkZp=C-0RQ@mail.gmail.com>
 <c9180ac41b00543e3531a343afae8f5bdca64d8d.camel@redhat.com>
 <20220907034407-mutt-send-email-mst@kernel.org> <d32101bb-783f-dbd1-545a-be291c27cb63@redhat.com>
 <20220908011858-mutt-send-email-mst@kernel.org> <c8cd9a2e-3480-6ca5-96fa-4b5bd2c1174a@redhat.com>
 <20221009160520-mutt-send-email-mst@kernel.org>
In-Reply-To: <20221009160520-mutt-send-email-mst@kernel.org>
From:   Jason Wang <jasowang@redhat.com>
Date:   Wed, 12 Oct 2022 11:19:12 +0800
Message-ID: <CACGkMEscu+mUBff1JUW4QxkyV33MwRP7VPSZ2-OXp5=pJaHC6Q@mail.gmail.com>
Subject: Re: [PATCH net] virtio-net: add cond_resched() to the command waiting loop
To:     "Michael S. Tsirkin" <mst@redhat.com>
Cc:     Jakub Kicinski <kuba@kernel.org>, Paolo Abeni <pabeni@redhat.com>,
        netdev <netdev@vger.kernel.org>,
        linux-kernel <linux-kernel@vger.kernel.org>,
        virtualization <virtualization@lists.linux-foundation.org>,
        Eric Dumazet <edumazet@google.com>,
        Gautam Dawar <gautam.dawar@xilinx.com>,
        davem <davem@davemloft.net>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Precedence: bulk

On Tue, Oct 11, 2022 at 1:11 AM Michael S. Tsirkin <mst@redhat.com> wrote:
>
> On Sun, Oct 09, 2022 at 01:58:53PM +0800, Jason Wang wrote:
> >
> > =E5=9C=A8 2022/9/8 13:19, Michael S. Tsirkin =E5=86=99=E9=81=93:
> > > On Thu, Sep 08, 2022 at 10:21:45AM +0800, Jason Wang wrote:
> > > > =E5=9C=A8 2022/9/7 15:46, Michael S. Tsirkin =E5=86=99=E9=81=93:
> > > > > On Wed, Sep 07, 2022 at 09:07:20AM +0200, Paolo Abeni wrote:
> > > > > > On Wed, 2022-09-07 at 10:09 +0800, Jason Wang wrote:
> > > > > > > On Tue, Sep 6, 2022 at 6:56 PM Paolo Abeni <pabeni@redhat.com=
> wrote:
> > > > > > > > On Mon, 2022-09-05 at 15:49 +0800, Jason Wang wrote:
> > > > > > > > > On Mon, Sep 5, 2022 at 3:15 PM Michael S. Tsirkin <mst@re=
dhat.com> wrote:
> > > > > > > > > > On Mon, Sep 05, 2022 at 12:53:41PM +0800, Jason Wang wr=
ote:
> > > > > > > > > > > Adding cond_resched() to the command waiting loop for=
 a better
> > > > > > > > > > > co-operation with the scheduler. This allows to give =
CPU a breath to
> > > > > > > > > > > run other task(workqueue) instead of busy looping whe=
n preemption is
> > > > > > > > > > > not allowed.
> > > > > > > > > > >
> > > > > > > > > > > What's more important. This is a must for some vDPA p=
arent to work
> > > > > > > > > > > since control virtqueue is emulated via a workqueue f=
or those parents.
> > > > > > > > > > >
> > > > > > > > > > > Fixes: bda324fd037a ("vdpasim: control virtqueue supp=
ort")
> > > > > > > > > > That's a weird commit to fix. so it fixes the simulator=
?
> > > > > > > > > Yes, since the simulator is using a workqueue to handle c=
ontrol virtueue.
> > > > > > > > Uhmm... touching a driver for a simulator's sake looks a li=
ttle weird.
> > > > > > > Simulator is not the only one that is using a workqueue (but =
should be
> > > > > > > the first).
> > > > > > >
> > > > > > > I can see  that the mlx5 vDPA driver is using a workqueue as =
well (see
> > > > > > > mlx5_vdpa_kick_vq()).
> > > > > > >
> > > > > > > And in the case of VDUSE, it needs to wait for the response f=
rom the
> > > > > > > userspace, this means cond_resched() is probably a must for t=
he case
> > > > > > > like UP.
> > > > > > >
> > > > > > > > Additionally, if the bug is vdpasim, I think it's better to=
 try to
> > > > > > > > solve it there, if possible.
> > > > > > > >
> > > > > > > > Looking at vdpasim_net_work() and vdpasim_blk_work() it loo=
ks like
> > > > > > > > neither needs a process context, so perhaps you could rewor=
k it to run
> > > > > > > > the work_fn() directly from vdpasim_kick_vq(), at least for=
 the control
> > > > > > > > virtqueue?
> > > > > > > It's possible (but require some rework on the simulator core)=
. But
> > > > > > > considering we have other similar use cases, it looks better =
to solve
> > > > > > > it in the virtio-net driver.
> > > > > > I see.
> > > > > >
> > > > > > > Additionally, this may have better behaviour when using for t=
he buggy
> > > > > > > hardware (e.g the control virtqueue takes too long to respond=
). We may
> > > > > > > consider switching to use interrupt/sleep in the future (but =
not
> > > > > > > suitable for -net).
> > > > > > Agreed. Possibly a timeout could be useful, too.
> > > > > >
> > > > > > Cheers,
> > > > > >
> > > > > > Paolo
> > > > > Hmm timeouts are kind of arbitrary.
> > > > > regular drivers basically derive them from hardware
> > > > > behaviour but with a generic driver like virtio it's harder.
> > > > > I guess we could add timeout as a config field, have
> > > > > device make a promise to the driver.
> > > > >
> > > > > Making the wait interruptible seems more reasonable.
> > > >
> > > > Yes, but I think we still need this patch for -net and -stable.
> > > >
> > > > Thanks
> > > I was referring to Paolo's idea of having a timeout.
> >
> >
> > Ok, I think we're fine with this patch. Any chance to merge this or do =
I
> > need to resend?
> >
> > Thanks
>
> Last question: do we want cpu_relax here now? Or is cond_resched
> sufficient?

(Have answered in another thread)

I think we need cpu_relax() since there could be no high priority task
in the current cpu so we still need to relax.

Thanks

>
> >
> > >
>